About Predict Where a Fire Will Spread: Sensors, Weather, and Fire Models

Moisture of Extinction is the Fuel Moisture Above Which Fire Stops Spreading Computational Fluid Dynamics Fire Behaviour Needs Midflame Wind, Not the Wind at Standard Measurement Height Dry Thunderstorms Ignite Fires While Their Rain Evaporates Before Reaching the Ground Polar-Orbiting Sensors Flag a Hot Pixel, Not a Fire Outline Active Crowning Needs Enough Canopy Bulk Density to Sustain the Crown Fire Fuel Treatment, Detection and Suppression Compete for the Same Mitigation Money Fire Radiative Power Turns a Hot Pixel Into an Energy Rate NFDRS Splits Available Energy from Expected Flame Length as ERC and Burning Index Linear Heat Detection Cable Makes an Entire Run the Sensor Radiation, Convection and Conduction Preheat the Fuel Ahead of a Fire Front A Red Flag Warning Asserts a Forecast Threshold, Not an Ignition Sensor Calibration The Ceiling Jet Is the Thin Fast Flow That Reaches Detectors First Spread Models Are Scored by Perimeter Overlap and Arrival-Time Error Multi-Criteria Detectors Weigh Several Fire Signatures Instead of One Threshold A Wind-Driven Fire Grows as an Ellipse with a Length-to-Breadth Ratio Operational Spread Models Are Tuned With Adjustment Factors, Not Trusted Raw False Positive Rate Computation Methods Heat Release Rate Is the Single Number That Sizes a Fire A Room Fire Becomes Ventilation-Controlled When the Opening Limits the Burning Rate A Raster Fire Automaton Ignites Neighbour Cells and Distorts the Fire Shape Wireless Sensor Network Architecture in Distributed Systems Detector Spacing Tables Are a Shortcut for a Ceiling Jet Calculation A Fire Plume Rises by Buoyancy and Cools by Entraining Air Fuel Bed Depth and Packing Ratio Decide How Air Reaches the Flames Burn Probability Comes from Simulating Thousands of Fires Over Sampled Weather The Rothermel Equation Divides Propagating Heat Flux by a Fuel Heat Sink Planning, Operations, Design and Research Each Need a Different Fire Model Metal-Oxide Gas Sensors Trade Sensitivity for Cross-Sensitivity and Drift Wind Forecast Error Usually Dominates Fire Spread Prediction Error Assimilating an Observed Perimeter Restarts the Forecast Where the Fire Actually Is The Canadian FWI System Chains Three Moisture Codes into Two Behaviour Indices A Wind Shift Turns a Fire's Long Flank Into a Wide Head Verification and Validation Ask Different Questions of a Fire Model A Near-Real-Time Hot-Spot Feed Carries Latency and Industrial False Alarms Monte Carlo Method Drought Indices Track Moisture in Fuels That a Single Rainfall Cannot Wet Video Smoke Detection Classifies Moving Texture Rather Than Particles Wind and Slope Multiply the No-Wind Rothermel Spread Rate Smouldering Smoke Carries Large Pale Particles and Flaming Smoke Small Dark Ones A Surface Fire Becomes a Crown Fire by Climbing Ladder Fuels Gusts and Wind Steadiness Set Variability a Spread Forecast Cannot Resolve A Fuel Ignites Once Incident Heat Flux Exceeds Its Critical Value Fuel Moisture is Estimated From Weather and Checked Against Fuel Sticks A Fire Danger Rating is a Relative Index, Not a Prediction of Fire Spread A Spread Forecast Inherits Every Error in Its Fuel Map Physics-Based Fire Models Solve Combustion in Cells and Cost Hours per Minute Vapor Pressure: Equilibrium Between Evaporation and Condensation in Liquids at Constant Temperature Flame Detectors Watch the Radiation Bands a Fire Uniquely Emits Surface-Area-to-Volume Ratio Separates Fine Flashy Fuels From Coarse Slow Ones Chimneys, Saddles and Narrow Canyons Channel Wind and Accelerate Fire Removing Any One Leg of the Fire Tetrahedron Stops Combustion Fire Model Fidelity Is Bought With Run Time, From Milliseconds to Days A Compartment Fire Builds a Hot Gas Layer Beneath the Ceiling Fire Models Range from Fitted Curves to Solved Conservation Equations The Australian FFDI Compresses Drought, Heat, Humidity and Wind into One Number Rotating Camera Towers Spot Smoke Plumes and Triangulate Their Origin Huygens Wavelet Propagation Grows a Perimeter by Spreading an Ellipse from Every Vertex Lightning Detection Networks Locate the Strikes That Start Fires Dew Point Tracks Real Moisture While Relative Humidity Swings with the Daily Temperature Cycle Ionization and Photoelectric Smoke Alarms Answer to Different Fires Operational Systems Like the Canadian FBP Predict Spread from Fitted Curves Head, Flanks and Heel Spread at Different Rates Around a Fire Perimeter Weather Station Observations Are the Input Path to Every Fire Danger Index Many Fires Grow With the Square of Time, From Slow to Ultrafast Dijkstra's Shortest Path Algorithm Implementation in Graph Theory Crown Fire Begins When Fireline Intensity Exceeds a Canopy Base Height Threshold Rate of Spread Measures How Fast a Fire Front Advances A Fire Spread Model Predicts Rate, Perimeter, Arrival Time and Intensity The Haines Index Rates the Lower Atmosphere's Potential for Erratic Fire Growth A Nighttime Inversion Holds a Fire Down Until It Breaks Up After Sunrise Water Mist and Gaseous Clean Agents Suppress by Different Mechanisms Detector Sensitivity Is Stated as Smoke Obscuration Per Unit Length Plume and Ceiling-Jet Correlations Predict the Hot Gas Reaching a Ceiling A Canopy Mesh of Gas Sensors Detects Fire Within Minutes Plume-Dominated Fires Are Driven by Their Own Convection, Wind-Driven Fires by the Wind Geostationary Imagers Revisit the Same Ground Every Few Minutes Spotting Starts New Fires Where Lofted Embers Land Ahead of the Front Drone Thermal Mapping Finds Hot Spots a Crew Cannot Reach Automated Weather Stations Are the Sensors Behind Fire Weather Numbers Conduction, Convection, and Radiation An Evacuation Trigger Point Is a Spread Model Run Backwards From Warning Time Thermal Cameras Choose Between Midwave and Longwave Infrared for Fire Piloted Ignition Needs a Spark While Autoignition Needs Only Enough Heat Level Set Representation of Moving Interfaces Aspect and Elevation Give Two Sides of a Ridge Different Fuel Moisture Flaming and Smouldering Are Two Combustion Regimes With Different Speeds and Signatures Foehn Windstorms Deliver the Driest and Fastest Fire Weather a Region Sees A Coupled Fire-Atmosphere Model Lets the Fire Rewrite Its Own Wind Field A Single Fire Perimeter Forecast Hides Uncertainty a Probability Map Shows Dead Fuel Timelag Classes Sort Fuels by How Fast They Track the Weather Fuel Moisture Content is Water Mass as a Percentage of Oven-Dry Fuel Mass A Zone Model Represents a Room Fire as Two Uniform Layers A Standard Fuel Model is a Parameter Set Standing In for Real Vegetation Fire Detection Latency, Coverage and Cost Cannot All Improve Together Cellular Automata as Discrete Dynamical Systems Kalman Filtering for Sensor Fusion and State Estimation Vapor Pressure Deficit Measures the Drying Power of Air on Dead Fuel The Cost of a Missed Smoulder Sets the Alarm Threshold Aspirating Detection Draws Air Through Pipes to a Laser Chamber A Level-Set Solver Moves the Fire Front as a Zero Contour Detector Sensitivity and Nuisance Alarm Rate Are One Setting, Not Two Sensitivity Analysis Finds the Single Number That Would Flip Your Decision Solid Fuels Burn Only After Pyrolysis Releases Flammable Gases The Cooking-Nuisance and Foam Fire Tests That Changed Alarm Design Relative Humidity Pyrocumulonimbus Forms When a Fire Plume Reaches Its Condensation Level Byram's Fireline Intensity Multiplies Heat Yield, Fuel Consumed and Rate of Spread Curing is the Fraction of a Grassland That Has Become Dead Fuel Carbon Monoxide Appears Before Visible Smoke in a Smouldering Fire Fuel Load Is the Dry Mass of Burnable Material per Unit Area A Spread Simulation Demands a Gridded Landscape of Fuel, Canopy and Terrain Atmospheric Stability Compares the Environmental Lapse Rate to a Rising Parcel's A Suppression System Waits for Confirmed Detection Before Releasing Airborne Infrared Line Scanners Map a Fire Perimeter Overnight Sensor Datasheet Interpretation A Learned Fire Model Fails Where Its Training Fires Never Went A Ceiling Detector Only Works Where the Smoke Actually Travels A Burn Probability Is a Frequency Over Simulated Seasons, Not a Forecast Heat of Combustion Sets the Energy a Kilogram of Fuel Can Release A Validated Fire Model Can Still Be Outside Its Domain of Applicability Fire Spreads Faster Uphill Because the Flame Leans Into Unburned Fuel Flashover Is the Moment Every Exposed Surface in a Room Ignites at Once Fixed-Temperature and Rate-of-Rise Heat Detectors Both Lag the Air Choose a Detector by the Fire It Must Catch and the Room It Watches A Machine-Learned Spread Model Fits Observed Fire Growth Instead of Physics Sensor Noise Sources Live Fuel Moisture is Set by Plant Water Status, Not by Yesterday's Weather Mixing Height and Transport Wind Describe the Air a Fire Has to Work With Sensor Transfer Function A Sprinkler Bulb's Response Time Index Predicts When It Opens Minimum Travel Time Solves Fire Arrival as a Shortest Path Across Cells Flame Length Is the Field-Observable Proxy for Fireline Intensity